C O R P O R A T E T E C H N O L O G Y MPLS Protection ...qofis.ccaba.upc.edu/pdf/CT1710.pdfvs. OSPF...

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003

Information & Communications

Networks &Multimedia

Communications

Simulative Comparison Simulative Comparison ofof

MPLS Protection SwitchingMPLS Protection Switchingvs. vs.

OSPF ReOSPF Re--routingrouting

Sandrine PASQUALINISandrine PASQUALINIAntoine FROTAntoine FROTAndreas Andreas IseltIselt

Andreas Andreas KirstädterKirstädter

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

CONTENTSCONTENTS•• IntroductionIntroduction

• MPLS

• Label switching

• Recovery

• OSPF

• Base protocol

• New routing algorithm

• Recovery

• Simulation

• Framework

• Scenarios

• Main results

• Conclusion

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Internet Internet EvolutionEvolution

• Issues :

• Increasing amount of multimedia flows over IP

• Need for Quality of Service (QoS)

• Objectives :

• Future and traditional services on a common IP-platform

• QoS, resilience, scalability

Network ControlServer

Network AdmissionControl

IP core

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

OurOur objectivesobjectives

• Simulate the behaviour of a routing protocol :

• Pre-calculated : MPLSMPLS

• Dynamic and distributed : OSPFOSPF

• Resilience : speed up failure detectionspeed up failure detection

improve reroutingimprove rerouting

• Compare the resulting recovery speed

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

CONTENTSCONTENTS• Introduction

•• MPLSMPLS

• Label switching

• Recovery

• OSPF

• Base protocol


• Recovery

• Simulation

• Framework

• Scenarios

• Main results

• Conclusion

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Label SwitchingLabel Switching

• Attribution of a label

• On ingress router (Label Edge Router)

• Not only destination based

• Suppression on egress router

• Packet forwarding

• On transient routers (Label Switch Routers)

• Following a Forwarding Table

• No analysis of packet’s network layer header

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

MPLS RecoveryMPLS RecoveryP

rote

ctio

nR

esto

rati

on

A

B C

DE F

G H I

P2: Regional Protection(by Haskin)

A

B C

DE F

G H I

P3: Local Protection

A

B C

DE F

G H I

R1: Global Restoration

A

B C

DE F

G H I

R3: Local Restoration

A

B C

DE F

G H I

R2: Regional Restoration

A

B C

DE F

G H I

P1: Global Protection

Global Regional Local

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications


• MPLS

• Label switching

• Recovery

•• OSPFOSPF

• Base protocol


• Recovery

• Simulation

• Framework

• Scenarios

• Main results

• Conclusion

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

OSPF OSPF –– Main mechanismsMain mechanisms

• Hello protocol

• Topology changes detection

• Link state advertisements (LSA)

• Distribution and storage of the topology

• Each router maintains its own view

• Routes calculation

• Distributed, shortest path

• Forwarding information base (FIB)

• Router’s architecture

• Updated when routes have changed

Thello 10s

Tdead 4* Thello

Tlsa 0.6 – 1.1ms

TlsaFlood 33ms

TSPT O(n.log n), O(n2)

TSPTdelay 5s

TSPThold 10s

TFIB 100 – 300ms

detectiondetection

propagationpropagation

reroutingrerouting

detectiondetection

propagationpropagation

reroutingrerouting

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Routes convergenceRoutes convergence

• Routes convergence time (RCT)

failure detection, information’s propagation, rerouting 40 40 –– 50s50s

• Possible extensions

• Reducing Thello, TSPTdelay et TSPThold

• React faster when resources decrease than when they increase :

• Multipath routing in order to have at least one backup link

• When a link fails, react only locally

expected RCT < 500ms< 500ms

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

""OutdegreeOutdegree 2" Routing (O2" Routing (O22))

• At least 2 outgoing links at each node for each destination

• No single point of failure

• No loops, except "jokers"

• Minimize: paths’ size/ number of jokers

• Today’s IP-network cores

are sufficiently meshed

• Example : COST 239

"hammoc set" towards Berlin

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Fault toleranceFault tolerance

• Speed up failure detection

• Reduce delays : Thello=100ms, TSPTdelay et TSPThold=0

• Detection at hardware level: 5ms

• Local reaction (no propagation delays)

• Change of routing tables

• Change of distribution weights

• Preventive mechanism

• After local reaction is completed, recompute routes

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications


• MPLS

• Label Switching

• Recovery

• OSPF

• Base protocol


• Recovery

•• SimulationSimulation

• Framework

• Scenarios

• Main results

• Conclusion

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Simulated protocolsSimulated protocols

• OSPF

• Standard

• Accelerated (Hello or hardware detection)

• Accelerated, with O2 routing and local reaction

• MPLS

• With a detection mechanism equivalent to OSPF Hellos

A

B C

DE F

G H I

P2: Regional Protection(by Haskin)

A

B C

DE F

G H I

P3: Local Protection

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Tool ChainTool ChainTopology

(.ndl)Demand

(.ndl)

Simulation(.tcl) NS-2

output(.nam) NAM

New routing algorithm

MPLS Paths (.ndl)

Pathmaker

Ndl2Tcl

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Simulation ScenariosSimulation Scenarios

• Characteristics

• COST 239 topology

• Fully-meshed traffic

• One link failure

• Measures

• Sum of all incoming traffic

• Mean of values for each possible link failure

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

User InterfaceUser Interface

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Comparison of SPF / OComparison of SPF / O2 2 routingrouting

• Just after the failure (2.5s)

• Overall traffic decreases

• Similar behaviour but

• More traffic is affected in O2 case

Overall traffic in the network [Mbytes]

Simulation time [s]

• With equal detection times

• Equal reaction times

O2

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Restoration timeRestoration time

100

1,20

33

0,8

0 20 40 60 80 100

Processing of a LSA

LSAflooding

Routescalculation

Routesinstallation

(ms)

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Comparison of SPF / OComparison of SPF / O2 2 routingrouting

The time saved thanks to OThe time saved thanks to O22

routing and local reaction isrouting and local reaction is

hidden by Thidden by TFIBFIB


Simulation time [s]

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Comparison of MPLS / OSPFComparison of MPLS / OSPF

• MPLS protection is always faster

• In general (curve a , standard MPLS)

• With equivalent detection times

(pairs a b , and c d )

100ms less than OSPF

TTFIBFIB


Simulation time [s]

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

ConclusionConclusion

• Resilience based on traditional OSPF is slow

• Accelerated OSPF :

• reaches sub-second recovery

• improving restoration speed ?

• speed up failure detection

• but FIB update still an issue

• MPLS protection has superior resilience :

• faster recovery

• but more complex operation, and more bandwidth consuming

Pre-planned LSP restoration with resource sharing

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Thank you for your attention

Questions ?

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Global Global ProtectionProtection

Protection switching, pre-established alternate LSP, global scopeProtection switching, pre-established alternate LSP, global scope

A E

GH

F

G

D

CB

p-LSP

w-LSP

+ Single backup LSP per working LSP– Failure signaling required+ Node failures covered

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Fast Fast RerouteReroute [[HaskinHaskin]]

u Alternative recovery LSP set up from the last-hop LSR in reverse direction to the ingress LSP and along a node-disjoint path to the egress LSP

Source: [draft-haskin-mpls-fast-reroute-01.txt]

Protection switching, pre-established, pre-reserved,

local switching, global recovery

Protection switching, pre-established, pre-reserved,


A E

GH

F

G

D

CB w-LSP

p-LSP

+ Single backup LSP per working LSP– No failure signaling required

+ Node failures covered– High spare capacity requirement

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

LocalLocal ProtectionProtection

Protection switching, pre-established, local scope, pre-reservedProtection switching, pre-established, local scope, pre-reserved

A E

GH

F

G

D

CB

w-LSP

p-LSP3

p-LSP2

p-LSP1

– Multiple backup LSPs per working LSP+ No failure signaling required– Node failures not covered

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

Global Global RestorationRestoration

Restoration, established on-demand, reserved on-demand, global scopeRestoration, established on-demand, reserved on-demand, global scope

A E

GH

F

G

D

CB

w-LSP

– Failure signaling required+ Node failures covered

+ Alternative LSPs distributed over network=> high spare capacity efficiency

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

FailureFailure to to EgressEgress RestorationRestoration

Restoration, pre-established, pre-reserved,


Restoration, pre-established, pre-reserved,


A E

GH

F

G

D

CB

w-LSP

+ No failure signaling required+ Node failures covered

o Between local and global routing=> average spare capacity efficiency

C O

R P

O R

A T

E

T E

C H

N O

L O

G Y

30 / 09 / 2003



Communications

LocalLocal RestorationRestoration

Restoration, established on-demand, reserved on-demand, local scopeRestoration, established on-demand, reserved on-demand, local scope

A E

GH

F

G

D

CB

w-LSP

+ No failure signaling required– Node failures difficult to cope with

– Alternative LSPs locally routed=> lower spare capacity efficiency

C O R P O R A T E T E C H N O L O G Y MPLS Protection ...qofis.ccaba.upc.edu/pdf/CT1710.pdfvs. OSPF...

Documents

Transcript of C O R P O R A T E T E C H N O L O G Y MPLS Protection ...qofis.ccaba.upc.edu/pdf/CT1710.pdfvs. OSPF...